What a Pattern Can, and Can't, Tell You
Let’s invent someone. Call her Priya. She has tracked her back pain for four months, and one evening, scrolling back, she sees it: the worst days cluster after the big weekly shop. Saturday shop, Sunday flare. Four of the last five weeks. She feels the small electric thrill of having found the thing.
Priya is fictional, and so is her pattern, which is why she is useful. Let’s follow her carefully, because what she does next is the whole difference between a record that helps and a record that misleads.
An association is a question, not an answer
What Priya has found is an association: two things that tend to arrive together. The National Library of Medicine, part of the National Institutes of Health, states in its guidance on health statistics the rule that applies to every diary ever kept: in the health sciences, finding a correlation between two variables is not enough, as correlation does not necessarily imply causality. The director of the National Center for Complementary and Integrative Health (NCCIH) made the same point about a study linking a herbal tea to longer life: even with statistical corrections, correlations don’t prove causality. The tea drinkers might simply have been the people who also ate better, moved more, and slept.
That is not a reason to ignore Priya’s Sundays. It is a reason to ask what kind of thing she has found. There are three candidates, and they look identical on the page.
Coincidence
Four of five is a small number. Priya has had roughly seventeen Saturdays in the record, and bad days scattered through it. Some of them were always going to land on Sundays. The way to check is to count the misses as carefully as the hits: the Saturdays she shopped and Sunday was fine, and the Sundays that were bad in weeks she did not shop. If the shop is followed by a bad day about as often as any other day is, she has found the human talent for seeing faces in clouds.
The record is good at this check, because it holds the boring weeks as faithfully as the exciting ones. Memory does not.
Timing
Say the count holds up: shop days really are followed by worse days more often than other days. The next question is which part of the shop. Priya’s Saturday contains a forty-minute drive, an hour on hard floors, carrying bags up two flights, and, because it is Saturday, a later night than usual. The association is with the day; the cause, if there is one, is with one thing inside it. Or two. Or the late night that has nothing to do with shopping at all.
There is also the matter of lag. Her back may not care about Saturday until Sunday afternoon, or Monday. Stress flares and post-exertional crashes both arrive late, and a same-day glance at the record misses them entirely. If the pattern is real, the lag is part of it, and worth naming: “the day after, not the day of.”
Something else driving both
The quiet possibility, and the one that catches most people. Priya shops on Saturdays because Saturday is when her week ends. Her week ends with a long day, a late night, and a Sunday of catching up on everything she postponed. Perhaps the shop is not the cause of the Sunday but a marker of it: the visible part of a week that is too full. Remove the shop and the Sundays may not change at all.
This is the NCCIH’s tea drinkers again. The thing you noticed is often standing in for the thing that matters, and a diary cannot tell the two apart on its own.
What Priya should do
Not stop shopping. Not decide it is the stairs. She should do the small things that turn an association into a better question:
- Count honestly. Hits and misses, over as many weeks as she has. If the pattern survives, keep going; if it does not, she has learned something true and saved herself a theory.
- Split the day. Note which parts of the shop happen each week. The weeks she has groceries delivered but still has the late night are the interesting ones.
- Name the lag. Same day, next day, two days. Write it down as part of the pattern.
- Test one harmless thing, if she can. Two weeks of delivery instead of carrying. Not as proof, as another few rows of evidence. For anything that touches medication or treatment, this step is a conversation with a clinician, not an experiment.
- Bring the whole thing as a question. “Bad days follow my Saturday shop about twice as often as other days, usually the next afternoon. I can’t tell if it’s the carrying, the standing, or the late night. Does that shape mean anything to you?”
That last sentence is the point of the record. It carries the observation, the count, the lag, and the uncertainty, and it hands the interpretation to someone with the training to do it. A clinician can do a great deal with a well-described pattern. They can do very little with a verdict.
What a pattern is for
A pattern in your record is a lead. Followed carefully, it narrows the question you bring to the next appointment, and it stops you spending a year avoiding something that was never the problem. Followed carelessly, it becomes a rule you live by that was only ever a coincidence with good timing. The difference is the count, the split, the lag, and the sentence that ends with a question mark.
Priya, being fictional, does all four. Her Saturdays turn out to be a late-night problem with a grocery-shaped shadow. That is invented too. Yours will be your own.
Where Flare fits
Flare - Chronic Pain Tracker is built to hold the boring weeks. Activities live in their own timeline columns, so “shop,” “stairs,” and “late night” can be logged separately and the split above becomes a glance. Trends Pro compares your ratings on days with and without each activity, stays quiet until there are enough days to compare, and puts the sample size on the card so a four-of-five pattern looks like what it is. Every insight in the app carries the same caveat this post does, on purpose: a pattern is a lead, not a verdict, and what it means is between you and your doctor.
Frequently asked questions
My symptom tracker found a pattern. Is it real?
Possibly. A pattern in a record is an association: two things that tend to arrive together. Whether one causes the other is a separate question, and a record on its own cannot answer it. Check whether the pattern repeats across many weeks, whether the order is consistent, and what else was true on those days. Then bring it to a clinician as a question, which is what it is.
What is the difference between correlation and causation in symptom tracking?
Correlation is "these happen together": bad nights and bad days, Thursdays and flares. Causation is "this makes that happen." The National Institutes of Health's statistics guidance puts it plainly: in the health sciences, finding a correlation between two variables is not enough, because correlation does not necessarily imply causality. A diary can show the first. Showing the second takes more than a diary.
How many times does a pattern need to repeat before I trust it?
There is no magic number, but a handful of matches out of a handful of chances is a coincidence waiting to happen. Look for the pattern across many occasions, and count the misses as carefully as the hits: the times the suspect happened and nothing followed, and the flares that arrived without it. A pattern that survives that count is worth raising.
Should I avoid a suspected trigger before talking to my doctor?
For something harmless, like a food or an activity you can easily do without for a couple of weeks, a careful test is reasonable and the diary is how you read it. For anything involving medication, treatment, or your general health, the pattern is a question for the clinician, not a change to make alone.
Sources
- Finding and Using Health Statistics: Causation — National Library of Medicine (NIH)
- Reduced Mortality Risks and Correlation vs. Causation — National Center for Complementary and Integrative Health (NIH)
Flare - Chronic Pain Tracker is a symptom-tracking tool, not a medical device. This article is general information, not medical advice — talk to your doctor about decisions affecting your health.